The Data Center Reckoning is Coming
One of AI's biggest cheerleaders just issued a warning: the data center boom is a trap. Alex Davis helped broker Nvidia's $20 billion Groq deal last week. Now he's telling investors to brace for a "significant financing crisis" in two years. As we close out 2025, even the true believers are questioning whether the build-it-and-they-will-come strategy will hold.
🏗️ AI's Biggest Cheerleader Warns of Data Center Bust
➡️ The move: Alex Davis, CEO of investment firm Disruptive and one of AI's most aggressive backers, warned investors in a year-end letter that speculative data center construction is heading for a crash. He predicts a "significant financing crisis" in 2027-2028 as third-party developers who borrowed heavily to build facilities discover that hyperscalers like Microsoft, Google, and Amazon plan to own their own infrastructure instead of leasing.
⚡ Why it matters: This warning comes from inside the house. Davis isn't some AI skeptic. His firm just facilitated Nvidia's largest deal ever. But he's seeing "too many business models with no realistic margin expansion story, extreme capex spend, lack of enterprise customer traction." The companies building AI infrastructure on spec may find themselves without tenants just as their debt comes due.
🎯 Your takeaway: The AI boom is real, but not every company riding the wave will survive. For professionals, this is a reminder that hype cycles have consequences. The companies that actually use AI to deliver value will outlast those building castles on speculation.
🔗 AI Skills Become an Open Standard Across Chatbots
➡️ The move: Anthropic launched Agent Skills as an open standard that lets AI agents execute specific tasks using natural language descriptions. Days later, OpenAI added support for Skills in ChatGPT Codex. Code sleuths have spotted evidence that Skills are coming to the main ChatGPT interface too, with the ability to convert custom GPTs into reusable skills.
⚡ Why it matters: For the first time, major AI companies are building on shared infrastructure. A skill you create for Claude could work with ChatGPT, Gemini, and Copilot. This interoperability means your investment in AI workflows becomes portable rather than locked to one platform.
🎯 Your takeaway: Pay attention to Skills as they roll out. The ability to package complex instructions into reusable, shareable components could make AI assistants genuinely useful for repetitive professional tasks, not just one-off conversations.
🎭 2025: The Year AI Got Its Vibe Check
➡️ The move: As 2025 closes, the industry is taking stock of a year that started with unprecedented exuberance and ended with harder questions. OpenAI raised $40 billion at a $300 billion valuation. Safe Superintelligence raised $2 billion before shipping a product. Hyperscalers promised $1.3 trillion in infrastructure spending. But the mood has shifted. Copyright lawsuits wound through courts. Reports of "AI psychosis" sparked calls for trust and safety reforms. Even Blue Owl Capital pulled out of a $10 billion Oracle data center deal.
⚡ Why it matters: The era of unquestioning AI enthusiasm is fading. Investors are asking whether AI companies can sustain their velocity. Does scaling require billions in the post-DeepSeek era? Is there a business model that returns even a fraction of the investment? These questions will define 2026.
🎯 Your takeaway: Healthy skepticism doesn't mean AI is a fad. It means the market is maturing. For professionals, this is good news. The pressure is shifting from "adopt AI now" to "adopt AI that actually works." Substance is starting to win over hype.
🔧 Tool Spotlight: MailMaestro
MailMaestro is an AI email assistant that learns to write like you, not just for you.
What makes it different: Most AI email tools generate generic professional copy. MailMaestro's "AI personality" feature trains on your preferences, including words to avoid, preferred tone, and job-specific context. Instead of spending hours training a general-purpose model on your writing style, MailMaestro handles the customization automatically.
Best for: Professionals who write dozens of emails daily and want consistency without the robotic tone. Particularly useful for sales, account management, and anyone who needs personalized communication at scale.
Pricing: Free plan with three requests per week. Paid plans start at $15/seat/month with a 14-day trial.
👉 Try it: Install from mailmaestro.ai and let it analyze a few of your sent emails to learn your voice.
✨ Try This Today: Learn Anything Faster with AI
New year, new skill to master? AI can accelerate your learning by acting as a personal tutor that adapts to exactly what you don't understand.
The technique: Use the Feynman method with AI. Explain a concept you're learning to your AI assistant as if teaching a curious friend. The AI will identify gaps in your understanding and probe deeper on weak spots.
How to use it:
- Choose something you're trying to learn (a software feature, a business concept, a new regulation)
- Prompt: "I'm going to explain [topic] to you. After I'm done, point out any gaps, misconceptions, or areas where I'm oversimplifying. Then quiz me on the parts I struggled with."
- Explain the concept in your own words
- Review the AI's feedback and answer its follow-up questions
- Repeat until you can explain it clearly and answer edge cases
Why it works: Teaching forces you to organize your knowledge. Having an infinitely patient tutor who catches your weak spots means you actually fill the gaps instead of glossing over them.
Time required: 10-15 minutes per concept. You'll retain more in that time than an hour of passive reading.
📚 Go deeper: https://neuralnotesedge.ai/learning-anything-faster/
Neural Notes
✨ The Wire
🔗 Oracle's AI debt is raising alarms as Barclays warns the company could face a cash crunch by November 2026, with its bonds now trading like junk despite official investment-grade ratings. Tom Tunguz
🔗 Google's A2UI project launched as an open-source tool that lets AI agents generate custom user interfaces on the fly based on conversation context, moving beyond text-only interactions. SD Times
🔗 AI-driven scientific breakthroughs accelerated in 2025, including Alzheimer's detection advances, humanoid robotics progress, and MIT using machine learning to find cement alternatives from 1 million rock samples. Axios
Neural Notes — AI that amplifies your value, not replaces it.
